Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for encoding an audio signal, the method comprising: generating a mapped envelope of a high band by mapping an envelope of the high band into a band configuration of a full band; generating an envelope of the full band by combining the mapped envelope of the high band with an envelope of a low band; generating bit allocation information for a sub-band based on the envelope of the full band; determining to perform envelope refinement if there is any sub-band to which a bit is allocated in the high band based on the bit allocation information; in response to determining to perform the envelope refinement, generating refinement data for the sub-band to which the bit is allocated in the high band, updating the mapped envelope by using the refinement data, updating the bit allocation information based on bits used for the envelope refinement for the sub-band to which the bit is allocated, and generating a bitstream including the refinement data.
2. The method of claim 1 , further comprising generating an excitation class based on signal characteristics of the high band and encoding the excitation class.
This invention relates to audio signal processing, specifically methods for encoding and decoding high-frequency audio signals in speech or audio codecs. The problem addressed is the efficient representation of high-band excitation signals, which are critical for maintaining audio quality but require significant data without proper compression. The method involves analyzing the high-band portion of an audio signal to extract signal characteristics, such as spectral shape, energy distribution, or temporal dynamics. These characteristics are used to generate an excitation class, which categorizes the high-band signal into distinct types or patterns. The excitation class is then encoded into a compact form, reducing the data required to represent the high-band signal while preserving perceptual quality. This encoded excitation class can later be used during decoding to reconstruct the high-band signal accurately. The method may also include generating a high-band excitation signal from a low-band excitation signal, where the low-band signal is modified or transformed to approximate the high-band characteristics. The excitation class helps guide this transformation, ensuring the synthesized high-band signal matches the original signal's properties. This approach is particularly useful in bandwidth extension techniques, where high-band information is reconstructed from a lower-bandwidth input. By encoding the excitation class, the method reduces the bitrate needed for high-band representation while maintaining audio fidelity, making it suitable for applications like voice communication, streaming, and storage-efficient audio encoding.
3. The method of claim 1 , wherein the updated bit allocation information is provided to be used for spectrum coding.
A method for updating bit allocation information in spectrum coding systems addresses the challenge of efficiently distributing available bits across frequency subbands to optimize signal transmission quality. The method involves dynamically adjusting bit allocation based on channel conditions, signal characteristics, or other system parameters to improve spectral efficiency and reduce distortion. The updated bit allocation information is then provided for use in spectrum coding, which involves encoding and decoding signals in the frequency domain to enhance performance in communication systems, audio processing, or other applications requiring efficient spectral representation. By adaptively allocating bits to different subbands, the method ensures that more bits are assigned to subbands with higher importance or sensitivity to quantization errors, while fewer bits are allocated to less critical subbands. This approach enhances overall system performance by balancing bit usage and minimizing perceptual or measurable distortion. The method is particularly useful in systems where spectral characteristics vary over time or where different subbands have varying significance, such as in adaptive audio coding, wireless communications, or multimedia transmission.
4. The method of claim 1 , wherein the generating of the refinement comprises calculating a delta of norm, which is a difference between the mapped envelope and an envelope from an original spectrum, by using a maximum limit and a minimum limit.
5. The method of claim 4 , wherein generating of the bitstream comprises generating the bitstream including necessary bits for representing the delta of norm and a value of the delta of norm.
This invention relates to video encoding, specifically improving bitstream efficiency by representing changes in normalization values (delta of norm) in a compressed format. The problem addressed is the inefficiency of transmitting full normalization values repeatedly, which consumes excessive bandwidth and processing resources. The solution involves encoding only the difference (delta) between consecutive normalization values, reducing the bitstream size while maintaining accuracy. The method first computes the delta of norm, which is the difference between a current normalization value and a previously encoded normalization value. This delta is then encoded into a bitstream using a compact representation. The bitstream includes both the necessary bits to represent the delta and the actual value of the delta. By transmitting only the delta rather than the full normalization value, the method minimizes redundancy and improves compression efficiency. The approach is particularly useful in video coding where normalization values are frequently updated, such as in transform coefficient quantization or adaptive filtering. By focusing on the delta rather than absolute values, the method reduces the computational overhead of encoding and decoding while maintaining signal fidelity. The technique can be integrated into existing video codecs to enhance performance without requiring significant architectural changes.
6. A method for decoding an audio signal, the method comprising: generating a mapped envelope of a high band by mapping an envelope of the high band into a band configuration of a full band; generating an envelope of the full band by combining the mapped envelope of the high band with an envelope of a low band; generating bit allocation information for a sub-band based on the envelope of the full band; determining to perform updating the envelope if there is any sub-band in which a bit is allocated in the high band based on the bit allocation information; and in response to determining to perform the updating the envelope, decoding refinement data for the sub-band to which the bit is allocated in the high band, and updating the envelope by using the refinement data, and updating the bit allocation information based on bits used for envelope refinement for the sub-band to which the bit is allocated.
7. The method of claim 6 , further comprising decoding an excitation class.
8. The method of claim 6 , the updated bit allocation information is provided to be used for spectrum decoding.
9. The method of claim 6 , wherein decoding of the refinement data comprises decoding necessary bits for representing a delta of norm and a value of the delta of norm, wherein the delta of norm is a difference between the mapped envelope and an envelope from an original spectrum.
10. An apparatus for encoding an audio signal, the apparatus comprising: at least one processor configured to: generate a mapped envelope of a high band by mapping an envelope of the high band into a band configuration of a full band; generate an envelope of the full band by combining the mapped envelope of the high band with an envelope of a low band; generate bit allocation information for a sub-band based on the envelope of the full band; determine to perform envelope refinement if there is any sub-band to which a bit is allocated in the high band based on the bit allocation information; in response to determining to perform the envelope refinement, generate refinement data for the sub-band to which the bit is allocated in the high band, update the mapped envelope by using the refinement data, update the bit allocation information based on bits used for the envelope refinement for the sub-band to which the bit is allocated, and generate a bitstream including the refinement data.
11. An apparatus for decoding an audio signal, the apparatus comprising: at least one processor configured to: generate a mapped envelope of a high band by mapping an envelope of the high band into a band configuration of a full band; generate an envelope of the full band by combining the mapped envelope of the high band with an envelope of a low band; generate bit allocation information for a sub-band based on the envelope of the full band; determine to perform updating the envelope if there is any sub-band in which a bit is allocated in the high band based on the bit allocation information; and in response to determining to perform the updating the envelope, decode refinement data for the sub-band to which the bit is allocated in the high band, update the envelope by using the refinement data, and update the bit allocation information based on bits used for envelope refinement for the sub-band to which the bit is allocated.
Unknown
February 2, 2021
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.